Mendeleev's Pattern of Valences
When the elements were listed in order of atomic weight, he found a sequence of valences that went
Wow--all the elements followed that pattern exactly?
Hydrogen had a valence of 1; helium hadn't been discovered yet. The elements
that became the second and third rows of the periodic table, beginning with
lithium (valence 1), followed the sequence perfectly. In the fourth row,
things became a bit messier, because the transition metals didn't seem
to fit the pattern--some of them even had more than one valence.
Mendeleev just put in the transition metals in order of atomic
weight, ignoring their valences; he found that the elements on either side
of them did follow the sequence:
Wait a second...if you ignore the transition metals, the periodic
table has eight columns, and there are only seven numbers in the sequence.
Excellent observation! Mendeleev's original table was missing a column. Helium, as I've said, was unknown to him, and it so happens that all the other elements in helium's group, the noble gases, were unknown as well; there was nothing to tell him he should leave gaps for them.
When some noble gases were discovered, chemists found that they didn't seem to react at all with other elements, so Mendeleev suggested that they should be given a valence of 0. Based on their atomic weights, he fit them into a column at the far right. The complete table, in terms of valences, looked like this: